Magnetic resonance imaging (MRI)-guided biopsy has emerged as an alternative to transrectal ultrasound (TRUS)-guided biopsy due to its ability to specifically target abnormal regions [1]. Several groups have developed MRI-compatible robotic devices to guide a biopsy needle accurately in an MRI [2-3]. These reports have been transrectal or transgluteal approaches and required moving the patient repeatedly in and out of the bore for imaging and needle insertion. We have reported the result of research and development of a robotic needle-guide device for in-bore transperineal biopsy [4]. The device is equipped with a 4-DOF manipulator that can position and orient a passive needle-guide or active needle insertion driver inside the bore. We present the initial results and the clinical feasibility of this manipulator using the passive needle-guide.

Subjects. The study was approved by the IRB. Total 13 men were enrolled including six men with prior negative TRUS biopsy and elevated prostate-specific antigen, six men on active surveillance (AS) for low-risk prostate cancer, and three men who were not candidates for TRUS-guided biopsy. Two men were excluded from evaluation as they had systematic non-target biopsy procedures.

Clinical Workflow. All procedures were performed in a 3T 70cm-bore MRI scanner (Siemens Verio) in the lithotomy position. The manipulator was composed of front and rear 2-DOF trapezoidal stages driven by piezoelectric actuators (Fig. 1). In this feasibility study, we kept the needle trajectory parallel to the B₀ field to simplify the planning and robot control. [Patient setup] A stabilizing frame was fixed on the tabletop (Fig. 2A). After covering the manipulator with a sterile drape, a sterile needle-guide was attached to stage. The manipulator was then docked on the tabletop (Fig. 2B). [Planning] Images of markers embedded in the frame and the prostate (planning images) were acquired. Navigation software (RadVision, Acoustic MedSystems) registered the robot to the image coordinate system by detecting the markers. Biopsy targets were defined on preprocedural multiparametric MRI by a single reader, mapped onto the planning images using non-rigid image registration in 3D Slicer software (https://www.slicer.org/), and imported to RadVision. [Needle placement] Once the needle-guide reached the designated location, the physician inserted an 18-gauge MRI-compatible core biopsy gun (Fully Automatic Biopsy Gun, InVivo Corporation) (Fig. 2C). [Needle adjustment and sampling] Images of the prostate with the needle were acquired (confirmation images). If the image revealed that the needle was not within the target lesion, reinsertion and reorientation was performed. Once the needle was confirmed to be in the target, the core biopsy sample was obtained. The total procedure time was recorded along with times for setup, planning, and biopsy (including needle placement, adjustment, and sampling).

Data Analysis. After reconstructing the needle trajectory on each confirmation image, the shortest distance from the planned target to the trajectory was computed as an accuracy metric. This metric has been used in previous clinical studies [2,6]. The distances were evaluated for first and best needle insertion attempts, where the best insertion was the attempt resulted in actual tissue sampling. The first insertion represents an accuracy achieved solely by the needle-guide, whereas the best insertion represents a realistic accuracy for sampling achieved by the aforementioned adjustment techniques.

All 11 biopsy procedures were successfully performed with an average of 1.8 targets per case. There were average 1.8 reinsertion attempts per target before taking the biopsy cores. The total time was 102.6±24.5 minutes (Table 1). The mean and standard deviation of targeting errors for the first and best attempts were 6.0±3.4 and 4.9±2.9 mm respectively. Pathology results confirmed prostate cancer in 10 men, with Gleason score ≥ 6 in five out of six men with previous negative biopsy, and five out of five men on AS. Two AS patients were upgraded to clinically significant cancers with Gleason score 7.In-bore MRI-guided prostate biopsy using the newly developed manipulator was feasible. The targeting errors were consistent with other clinical studies on MRI-guided biopsies ranging 5.7-10 mm [2,5,6]. The procedure times were shorter than our previous study on manual and robotic needle-guide template [6], but longer than the study on transrectal robotic biopsy with similar number of targets largely due to the long planning time in our study (25.6 minutes); this was in part related to a separate research study on intraprocedural prostate image registration. In this study, we did not take advantage of the full 4-DOF needle angulation capability to limit the complexity of the procedure. The 4-DOF placement would offer more options for trajectories, and thus it could potentially reduce the number of attempts.